Easing Obstacles to Teaching Synthetic Aperture Radar (SAR) Data Science: ASF OpenSARlab

The Alaska Satellite Facility (ASF) has developed the OpenSARlab service to ease the difficulties that students and scientists typically encounter as they learn how to access and work with Synthetic Aperture Radar (SAR) data. SAR data are large, making them expensive to download and store. Once a scientist has their data, working with them requires installing complicated computing environments, which can differ and conflict with each other depending on the software requirements for a given workflow.

For individual users, ASF's OpenSARlab provides free, limited access to a cloud-hosted Jupyter Hub that sits alongside ASF data archives in AWS. Transferring SAR data to user volumes is incredibly fast and comes at no cost to the user. ASF provides a library of data recipes in the form of Jupyter Notebooks, allowing users to explore a variety of SAR data analysis techniques. ASF provides extendable software environments (via Conda) that allows users to skip what is often a painful setup process, and start working with data right away. Each user also has access to a persistent 500GB volume, on which they can store data indefinitely.

For research teams, classes, or training sessions, ASF provides custom OpenSARlab deployments that enable full-performance SAR processing capabilities easily accessible through a web browser. These furnish all the benefits described above but can be further customized to offer the exact computational resources and software environments along with full costing control and the tailored Jupyter Notebooks needed for a given use-case.

ASF has used OpenSARlab to host a growing number of educational events, leveraging both SAR and non-SAR datasets. OpenSARlab is ideal for remote classes and training sessions because it can scale to accommodate large numbers of simultaneous users and is accessible from anywhere in the world via a web browser. Each user has the same processing environment, so educational time is not wasted debugging software installations, and final processing results are the same for all users.